Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/60992
Title: Differential evolution-based optimal Gabor filter model for fabric inspection
Authors: Tong, L
Wong, WK 
Kwong, CK 
Keywords: Defect detection
Differential evolution
Fabric quality inspection
Optimal Gabor filters
Issue Date: 2016
Publisher: Elsevier
Source: Neurocomputing, 2016, v. 173, p. 1386-1401 How to cite?
Journal: Neurocomputing 
Abstract: In this paper, a defect detection model using optimized Gabor filters, which is suitable for real-time operation, is proposed to tackle the woven fabric inspection problem in fashion industry. Based on the analysis of the particular characteristics of fabric defects, the proposed model utilizes composite differential evolution (CoDE) to optimize the parameters of Gabor filters, which can achieve the optimal feature extraction of fabric defects. Together with thresholding and fusion operations, the optimal Gabor filters can successfully segment the defects from the original image background. By using optimal Gabor filters instead of a Gabor filter bank, the computational cost of the detection model can be significantly reduced. The performance of the proposed defect detection model is evaluated off-line through extensive experiments based on various types of fabric. Experimental results reveal that the proposed detection model is effective and robust, and is superior than four existing models in terms of the high detection rate and low false alarm rate.
URI: http://hdl.handle.net/10397/60992
ISSN: 0925-2312
EISSN: 1872-8286
DOI: 10.1016/j.neucom.2015.09.011
Appears in Collections:Journal/Magazine Article

Access
View full-text via PolyU eLinks SFX Query
Show full item record

SCOPUSTM   
Citations

6
Last Week
0
Last month
Citations as of Aug 15, 2017

WEB OF SCIENCETM
Citations

7
Last Week
0
Last month
Citations as of Aug 12, 2017

Page view(s)

31
Last Week
2
Last month
Checked on Aug 13, 2017

Google ScholarTM

Check

Altmetric



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.